-pra currently says
This document describes an algorithm that allows one to determine who
appears to have most recently caused an e-mail message to be
delivered. It does this by inspecting the headers in the message.
[RFC2822] contains a detailed specification of all of the relevant
headers.
I think this is the wrong approach, particularly now we have split out
PRA from Sender ID, and it might be used for other things, too.
We should be defining an identity. Then we can comment on what it's
useful for.
I propose this should be replaced with:
This document defines a new identity associated with a message,
called the purported responsible address, which is determined by
inspecting the header of the message. [RFC2822] contains a
detailed specification of all of the relevant header fields.
The purported responsible address is designed to correspond to the
entity that (according to the header) most recently caused an
e-mail message to be delivered.
The algorithm for determining the purported responsible address
from the header of a message is given in section 2.
Also:
For this reason, the result of the algorithm is called the
"Purported Responsible Address" -- "purported" because it tells you
what a message claims about where it came from, but not necessarily
where it actually came from.
I believe this paragraph should be deleted as it relates to one
specific use of the PRA algorithm.
The PRA algorithm may be of use in circumstances where the headers are
known to be authentic. Indeed, [SUBMITTER] requires that the PRA
algorithm be run by the sender as well as the recipient.
(See also my separate post suggesting that we rename the purported
responsible address so as to remove the word "purported")
-roy